A transcriptional regulator linking quorum sensing and chitin induction to render Vibrio cholerae naturally transformable

(1)

A transcriptional regulator linking quorum sensing

and chitin induction to render Vibrio cholerae

naturally transformable

Mirella Lo Scrudato and Melanie Blokesch*

Global Health Institute, School of Life Sciences, Ecole Polytechnique Fe´de´rale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland

Received December 15, 2012; Revised January 6, 2013; Accepted January 8, 2013

ABSTRACT

The human pathogen Vibrio cholerae is an aquatic bacterium associated with zooplankton and their

chitinous exoskeletons. On chitinous surfaces,

V. cholerae initiates a developmental programme, known as natural competence, to mediate trans-formation, which is a mode of horizontal gene transfer. Competence facilitates the uptake of free DNA and recombination into the bacterial genome. Recent studies have indicated that chitin surfaces are required, but not sufficient to induce compe-tence. Two additional regulatory pathways, i.e. ca-tabolite repression and quorum sensing (QS), are

components of the regulatory network that

controls natural competence in V. cholerae. In this study, we investigated the link between chitin induc-tion and QS. We show that the major regulators of these two pathways, TfoX and HapR, are both involved in the activation of a gene encoding a tran-scriptional regulator of the LuxR-type family, which we named QS and TfoX-dependent regulator (QstR). We demonstrate that HapR binds the promoter of qstR in a site-specific manner, indicating a role for HapR as an activator of qstR. In addition, epistasis experiments indicate that QstR compensates for the absence of HapR. We also provide evidence that QstR is required for the proper expression of a small but essential subset of competence genes and propose a new regulatory model in which QstR links chitin-induced TfoX activity with QS. INTRODUCTION

The bacterium Vibrio cholerae is a facultative pathogen and the causative agent of cholera. Cholera is far from extinction and is even considered a re-emerging disease (1). V. cholerae commonly occurs in aquatic ecosystems,

its true habitat, where it intimately associates with zoo-plankton and their chitinous exoskeletons. Chitin induces natural competence for transformation in V. cholerae (2), a mode of horizontal gene transfer. In this state, the bac-terium can import and recombine DNA from the

environ-ment, thereby becoming naturally transformed.

Chitin-induced natural competence is not only speciﬁc for V. cholerae but is also conserved in other species of

the genus Vibrio and has been experimentally

demonstrated for Vibrio ﬁscheri, Vibrio vulniﬁcus and Vibrio parahaemolyticus (3–5).

Recent studies have demonstrated that there is a strong link between natural competence/transformation and the environmental niche of the bacterium (2,6–9). More spe-ciﬁcally, it was shown that chitin sensing and degradation, quorum sensing (QS) and carbon catabolite repression contribute to the onset of competence [for a recent review, see (10)]. Nevertheless, how these pathways are interconnected with respect to competence induction and natural transformation remains still poorly understood. Here, we describe a regulatory protein, which we named QS and TfoX-dependent regulator (QstR), as an inter-mediate regulator for natural competence induction and transformation, thereby linking chitin induction and QS (Figure 1).

The induction of tfoX, which encodes the major regula-tor of transformation in V. cholerae, in the presence of chitin was first demonstrated using microarray expression profiling (14). Indeed, tfoX was significantly upregulated upon the growth of V. cholerae on crab shell surfaces or, alternatively, in liquid cultures supplemented with N-acetylglucosamine oligomers (n > 2), but not on supple-mentation with the N-acetylglucosamine monomer (14). In 2005, experiments demonstrated for the first time that chitin renders V. cholerae naturally transformable and that this phenotype is fully dependent on TfoX (2). The authors of that previous study also showed that tfoX overexpression is sufficient to render V. cholerae naturally transformable, even in the absence of chitin as an inducer (2) (Figure 1, chitin independent). Subsequent studies have

*To whom correspondence should be addressed. Tel: +41 21 693 0653; Fax: +41 21 693 7210; Email: melanie.blokesch@epﬂ.ch ß The Author(s) 2013. Published by Oxford University Press.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/ by-nc/3.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

(2)

confirmed the requirement of chitin oligomers for tfoX induction (15). Furthermore, Yamamoto et al. (16) provided evidence for the involvement of a small regula-tory RNA, TfoR, which acts as an activator of tfoX trans-lation upon chitin induction [reviewed in (10)]. How the regulatory protein TfoX acts on downstream genes remains unknown. However, as the secondary messenger Adenosine 30_,50_{-cyclic monophosphate (cyclic AMP or} cAMP) and its receptor protein CRP are also crucial for natural competence and transformation of V. cholerae (9), the current idea with respect to TfoX-mediated compe-tence induction is based on a model proposed by Redfield for another naturally competent bacterium, Haemophilus influenzae (17–19) (Figure 1). In this organism, the TfoX-homolog Sxy is required for a CRP-cAMP-dependent induction of the ‘Sxy-dependent cyclic AMP receptor [CRP-S] regulon’ (20).

The third pathway that is crucial for natural competence and transformation of V. cholerae is QS (2,6–8,21–23). Bassler and collaborators have extensively studied QS in V. choleraefor many years [for a recent review, see (11)]. These studies have indicated that the regulatory circuitry of QS is incredibly complex, as it includes at least two different autoinducer molecules, i.e. cholera autoinducer 1 (CAI-1) and autoinducer 2 (AI-2) (24–26), receptor proteins acting as kinases/phosphatases, small regulatory RNAs (27) and many other regulatory elements (11,28).

Blokesch and Schoolnik demonstrated one mechanism underlying the natural transformation-negative phenotype of QS defective V. cholerae strains and the interconnection between QS and natural competence and transformation (7). These authors showed that on increased cell density, the major regulator of QS, HapR, represses (directly or indirectly) the gene dns, which encodes an extracellular nuclease (7) (Figure 1). As this nuclease degrades surround-ing DNA [(7), and recently conﬁrmed in (29)] that could potentially act as transforming material, its repression is crucial for natural transformation. Furthermore, based on previous microarray expression data (2), it was also speculated that HapR acts as an activator for the expres-sion of the essential competence gene comEA (7,21). ComEA is predicted to encode a periplasmic DNA-binding protein, which is involved in DNA uptake (2,7,21). The reduced expression of comEA in QS mutants was later conﬁrmed experimentally (8,22).

Lo Scrudato and Blokesch largely extended our view on the connection between QS and natural competence and transformation, providing evidence that HapR and Dns are inversely correlated at the protein level and demon-strating that QS and TfoX activity co-regulate only a minority of competence genes (8). Speciﬁcally, this study indicated that apart from comEA, only one other tested competence gene, comEC, also requires HapR for TfoX-dependent expression (8). ComEA, which is most

Figure 1. Schematic representation of the regulatory circuitry of natural competence and transformation of V. cholerae. Upon growth on chitin surfaces (or chitin-independent artifical induction), the expression of tfoX, encoding for the main regulator of transformation TfoX, occurs. Concomitantly with cAMP binding to CRP, TfoX most likely induces the expression of the competence genes, which include the genes encoding the assembly machinery and structural components of a type IV pilus (pil genes) in V. cholerae. TfoX also positively regulates chitin metabolism genes, such as those encoding chitinases (chiA-1 and chiA-2 depicted as chiA in the scheme). In this study, we provided evidence for the existence of an intermediate transcription factor downstream of TfoX, QstR, which is required for the expression of a small subset of competence genes (comEA and comEC). We showed that the expression of these genes, which are also dependent on the QS circuitry, is mediated through QstR, which itself is dependent on the main regulator of QS, HapR. QstR thus links the TfoX- and QS-dependent signalling in V. cholerae. At this point, an additional regulation of comEA/comEC by TfoX/CRP-cAMP cannot be excluded and is indicated by the grey dashed arrow. HapR is primarily produced in the presence of high levels of the CAI-1, (whereas AI-2 only plays a minor role in the production of HapR) (8), reflecting the high cell density (HCD) of the population (11). Earlier studies have demonstrated that HapR binds to the promoter sequences of the two competence-unrelated genes (aphA and hapA) (12,13), which we used as controls in this study. Here, we identified putative HapR binding sites upstream of qstR and dns (black boxes) based on the in vitro binding of HapR to these promoter regions and previous in silico predictions (grey boxes) (13).

(3)

likely a periplasmic DNA-binding protein (2), and ComEC, which encodes an inner membrane transporter (21), play a major role in the DNA uptake process, as both of them are supposed to directly interact with the incoming DNA (23). In contrast, the expression of the competence genes with a potential role in scaffolding or type IV-like pilus assembly [e.g., pilA and pilM; (2,21,30)] was QS independent (8). Interestingly, Suckow et al. (21) and Lo Scrudato and Blokesch (8) unambiguously showed that natural transformation was almost exclusively de-pendent on the major autoinducer of V. cholerae, CAI-1. These data were consistent with the previous results of Zhu and Mekalanos (31) who investigated QS-dependent biofilm formation in V. cholerae. These authors wrote, ‘AI-2 signals are largely dispensable, while CAI-1 sig-nalling is important for regulating biofilm formation’ (31). The extremely minor contribution of AI-2 to natural transformation was also reflected in another recent study (22). The data of this study showed that natural transformation was also highly reduced in a CAI-1 negative strain, whereas the transformation frequencies did not drop significantly below that level in a strain lacking both autoinducers (e.g. CAI-1 and AI-2 negative) (22). Furthermore, natural transformation occurred even in the absence of both autoinducers, which was not observed in our previous studies (8,21). Based on the strong connection between CAI-1 and natural competence/transformation, we proposed that CAI-1 acts as a competence pheromone (8,21,23). Notably, another gram-negative and naturally competent bacterium, Legionella pneumophila (32), also produces a a-hydroxyketone (AHK) autoinducer called Legionella autoinducer 1 [LAI-1, a 3-hydroxypentadecane-4-one; (33)], which is similar to CAI-1 [a 3-hydroxytridecane-4-one; (26)] [for a recent review on AHK, see (34)]. Based on this and the hypothesis that CAI-1 is a com-petence pheromone (8,21), Seitz and Blokesch (10) recently wrote that ‘it is tempting to speculate that a-hydroxyketone signalling molecules are commonly involved in the regulation of natural competence’. Indeed, Kessler et al. (35) recently showed that LAI-1 and its respective sensor kinases play a major role in the natural competence of L. pneumophila, although in a re-ciprocal manner from that of V. cholerae. But even though a lot of information is available concerning the upstream regulatory circuitry of the major regulator of QS, HapR, and its contribution to natural competence and transform-ation, the downstream signalling pathway has not been experimentally challenged. This report is the first to dem-onstrate that HapR directly binds the promoter region of dns, but not that of comEA. In contrast, HapR binds the promoter region and activates the expression of a gene encoding an intermediate regulatory protein, QstR. The expression of qstR also requires TfoX, apart from HapR, thus linking QS and (chitin-induced) TfoX induc-tion in V. cholerae. Moreover, we demonstrate that the artificial expression of qstR restores natural competence and transformation in hapR-deficient strains, further con-firming a role for QstR as an intermediate regulator in the natural competence and transformation of V. cholerae.

MATERIALS AND METHODS Media and growth conditions

Vibrio choleraeand Escherichia coli strains were grown in Luria both (LB) medium at 30_{C with shaking, unless} otherwise stated. Antibiotics were added for plasmid maintenance or transformants/transconjugants selection at concentrations of 50 or 100 mg/ml for ampicillin, 75 mg/ml for kanamycin and at 50 mg/ml for gentamicin. Thiosulfate-citrate-bile salts-sucrose agar plates were used to counter-select E. coli after bi-/triparental mating with V. cholerae; the thiosulfate-citrate-bile salts-sucrose agar plates were prepared according to the manufacturer’s in-structions (Sigma-Aldrich/Fluka, Buchs, Switzerland). NaCl-free LB medium containing 6% sucrose was prepared for sucrose-based sacB-counter-selection. Bacterial strains and plasmids

The V. cholerae strains and plasmids used in this study are indicated in Table 1. The E. coli strains DH5a (43) and S17-1pir (44) were used for cloning and as a donor strain in bacterial mating experiments, respectively. E. coli Origami 2TM (DE3) pLysS (Novagen) was used as a host for HapR protein expression.

Construction of V. cholerae mutant strains

The gene qstR (VC0396) was deleted from the parental strain A1552 using the gene disruption method based on the counter-selectable plasmid pGP704-Sac28, as previously described (14). The oligonucleotides used to construct the respective plasmid are indicated in Supplementary Table S1.

Construction of plasmids used to investigate the putative HapR binding site upstream comEA

Most of the plasmids used in this study were derived from the plasmid pBR322 (39) (Table 1). Initially, pBR322 was modiﬁed through the partial deletion of the tetracycline resistance cassette and the constitutive promoter PTet,

re-sulting in the plasmid pBR-Tet_MCSII (Table 1). The primers P[VC1917]-GFP#1-NotI and VC1917-down-NotI (Supplementary Table S1) were used in polymerase chain reaction (PCR) to amplify comEA preceded by 900 bp of its upstream region. The genomic DNA (gDNA) from the V. choleraestrain A1552 was used as a template. The PCR fragment was digested with NotI and cloned into the equally digested vector pBR-Tet_MCSII, resulting in the plasmid pBR-[own]comEA.

The plasmid pBR-[own]comEA was digested with AatII and BglII to shorten the upstream region of comEA. Self-ligation resulted in the plasmid pBR-[Pwt]comEA,

carrying comEA and 200 bp of its upstream region (Table 1). Site-directed mutagenesis using inverse PCR on plasmid pBR-[Pwt]comEA generated the plasmids

pBR-[Pmut L]comEA, pBR-[Pmut R]comEA and

pBR-[Pmut L/R]comEA, carrying mutations in the putative

HapR binding site upstream comEA (see oligonucleotides in Supplementary Table S1).

(4)

Construction of complementing plasmids and plasmid pBAD-hapR-N-Strep

The genes hapR, comEA and qstR were ampliﬁed with the respective primers indicated in Supplementary Table S1 using gDNA from V. cholerae strain A1552 as a template. The NcoI and EcoRI-digested PCR products were cloned into the equally digested plasmid pBAD/ Myc-HisA to generate plasmids hapR, pBAD-comEA and pBAD-qstR (Table 1). The Strep-tagIIÕ encoding sequence was added to hapR through inverse PCR using the oligonucleotides hapR-N-Strep-fw and hapR-N-Strep-bw (Supplementary Table S1) and pBAD-hapRas a template, yielding plasmid pBAD-hapR-N-Strep.

Construction of plasmids containing the putative HapR binding sites upstream of dns and qstR

The short DNA segments upstream of dns and qstR, re-spectively (30–50 bp), were introduced as overhangs in the oligonucleotides indicated in Supplementary Table S1. The primers were used to amplify pBR-Tet_MCSI through inverse PCR to generate the plasmids pBR-[Pdns-100-50], pBR-[Pdns-50-1], pBR-[PqstR-150-120] and

pBR-[PqstR-150-102] (Table 1). For the electrophoretic

mobility shift assay (EMSA) experiments, the respective inserted DNA regions, ﬂanked by DNA sequences derived from the plasmid (200-bp fragments), were PCR ampliﬁed using the primers pBR-TET_MCS-before and

Table 1. Bacterial strains and plasmids

Strains or plasmids Genotypea _Reference

Strains (V. cholerae)

A1552 Wild-type (WT), O1 El Tor Inaba, RifR (36) A1552-LacZ-Kan A1552 strain with aph cassette in lacZ gene; RifR_{, Kan}R _(37,38)

A1552-TntfoX A1552 containing mini-Tn7-araC-PBAD-tfoX; RifR, GentR (8)

hapR A1552VC0583, RifR (2) hapR-TntfoX A1552hapR containing mini-Tn7-araC-PBAD-tfoX; RifR, GentR (8)

comEA A1552VC1917 [=A1552VC1917 in (2)], RifR (2) comEA-TntfoX A1552comEA containing mini-Tn7-araC-PBAD-tfoX; RifR, GentR This study

qstR A1552VC0396, RifR This study qstR-TntfoX A1552qstR containing mini-Tn7-araC-PBAD-tfoX; RifR, GentR This study

Plasmids

pBR322 AmpR_{, Tc}R ₍₃₉₎

pGP704-Sac28 Suicide vector, ori R6K sacB, AmpR ₍₁₄₎

pGP704-28-SacB-qstR pGP704-Sac28 with a gene fragment resulting in a 402-bp deletion (incl. stop codon) within VC0396 (qstR)

This study pBAD/Myc-HisA pBR322-derived expression vector; araBAD promoter (PBAD); Amp

R

Invitrogen pBAD-comEA comEAgene cloned into pBAD/Myc-HisA, arabinose inducible; AmpR _{This study}

pBAD-hapR hapRgene cloned into pBAD/Myc-HisA, arabinose inducible; AmpR This study pBAD-hapR-N-Strep hapRgene preceded by sequence encoding Strep-tagIIÕ _{cloned into pBAD/}

Myc-HisA, arabinose inducible; AmpR

This study pBAD-qstR qstRgene cloned into pBAD/Myc-HisA, arabinose inducible; AmpR _{This study}

pUX-BF13 oriR6K, helper plasmid with Tn7 transposition function; AmpR ₍₄₀₎

pGP704::Tn7 pGP704 with mini-Tn7 Schoolnik laboratory collec-tion; (41)

pGP704-mTn7-araC-tfoX

pGP704 with mini-Tn7 carrying araC and PBAD-driven tfoX; Amp R

(8) pBR-Tet_MCSI pBR322 derivative deleted for Tet promoter and part of tetRgene; AmpR (8) pBR-Tet_MCSII pBR322 derivative deleted for Tet promoter and part of tetR_{gene; new MCS}

included; AmpR

This study pBR-[own]comEA comEAgene preceded by 900 bp of upstream region cloned into

pBR-Tet_MCSII; AmpR

This study pBR-[PWT]comEA comEAgene preceded by 200 bp of upstream region; AmpR This study

pBR-[Pmut L]comEA plasmid generated by inverse PCR of pBR-[PWT]comEA; mutated within the

left part of the putative HapR binding site upstream comEA; AmpR

This study pBR-[Pmut R]comEA plasmid generated by inverse PCR of pBR-[PWT]comEA; mutated within the

right part of the putative HapR binding site upstream comEA; AmpR

This study pBR-[Pmut L/R]comEA plasmid generated by inverse PCR of pBR-[PWT]comEA; mutated within the

left and right part of the putative HapR binding site upstream comEA; AmpR

This study

pBR-[Pdns-100-50] DNA sequence corresponding to the 100 to 50 bp region upstream of dns

was inserted into pBR-Tet_MCSI by inverse PCR; AmpR This study

pBR-[Pdns-50-1] DNA sequence corresponding to the 50 to 1 bp region upstream of dns

was inserted into pBR-Tet_MCSI by inverse PCR; AmpR

This study pBR-[PqstR-150-120] DNA sequence corresponding to the 150 to 120 bp region upstream of

qstRwas inserted into pBR-Tet_MCSI by inverse PCR; AmpR

This study pBR-[PqstR-150-102] DNA sequence corresponding to the 150 to 102 bp region upstream of

qstRwas inserted into pBR-Tet_MCSI by inverse PCR; AmpR

This study pBR-[PqstR

-150-102]_mut_AgeI

plasmid generated by inverse PCR of pBR- [PqstR-150–102]; site-directly

mutated in two bases as indicated in Figure 6D thereby creating an AgeI re-striction enzyme recognition site; AmpR

This study

(5)

pBR-TET_MCS-after (Supplementary Table S1). The plasmid pBR-[PqstR-150-102]_mut_AgeI, carrying

muta-tions in the promoter region 150 to 102 upstream of the start codon of qstR (PqstR; see Figure 6D), was

con-structed through inverse PCR using the oligonucleotides

PVC0396_AgeI_fw and PVC0396_AgeI_bw and the

plasmid pBR-[PqstR-150-102] as a template.

Natural transformation assay (chitin-dependent and chitin-independent)

Chitin-dependent natural transformation assays were done on chitin ﬂakes as previously described by Marvig and Blokesch (37,45). Chitin-independent induction of natural competence and scoring of transformants was per-formed using V. cholerae strains carrying an inducible chromosomal copy of TfoX as described by Lo Scrudato and Blokesch (8). Statistical analyses of transformation data were done on log-transformed data (46) using a two-tailed Student’s t-test.

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis and western blotting

The proteins were separated through sodium dodecyl sulfate–polyacrylamide gel electrophoresis and then either stained for protein using Coomassie blue or subjected to western blotting as previously described (8). Primary antibodies against HapR (8) and against the Strep-tagII sequence (a-Strep-MAB classic, IBA GmbH, Go¨ttingen, Germany) were diluted at 1:10 000 and 1:1000, respectively. Goat anti-rabbit horseradish peroxidase (HRP) (diluted 1:20 000; Sigma-Aldrich, Switzerland) and goat anti-mouse HRP (diluted 1:5000; Sigma-Aldrich, Switzerland) were used as secondary antibodies. Lumi-LightPLUS western blotting substrate (Roche, Rotkreuz, Switzerland) or Western LightningÕ-ECL (PerkinElmer, Schwerzenbach, Switzerland) were used as HRP substrates, and the lumi-nescence signals were detected using chemilumilumi-nescence-

chemiluminescence-detecting ﬁlms (Amersham Hyperﬁlm ECL, GE

Healthcare via VWR, Dietikon, Switzerland). Puriﬁcation of HapR-N-Strep

HapR-N-Strep protein was expressed in the E. coli strain OrigamiTM 2 (DE3) pLysS (Novagen) carrying the plasmid pBAD-hapR-N-Strep. The cells were grown in LB medium at 30_{C until the culture reached an optical} density of 0.8 at 600 nm. At that time, the expression was induced on the addition of 0.2% arabinose to the culture medium. After induction, the bacteria were further cultivated at 16_{C overnight. The cells were collected} through centrifugation and resuspended in lysis buffer (50 mM sodium phosphate pH 8, 125 mM NaCl and 1% Triton) containing a protease inhibitor cocktail (Roche, Rotkreuz, Switzerland). The cells were disrupted using several passages through a French press cell. The crude extract was clariﬁed through centrifugation (17 000 rpm for 30 min at 4_{C), and the supernatant was loaded onto} two individual Strep-TactinÕ _{Sepharose columns (1 ml of} column volume; IBA GmbH, Go¨ttingen, Germany). Each column was washed with 5 volumes of washing buffer (100 mM Tris–HCl, pH 8.0, 150 mM NaCl and 1 mM

ethylenediaminetetraacetic acid) to remove the unbound proteins. The protein was eluted in six fractions (each at 0.5 column volume) of the same buffer containing 2.5 mM D-desthiobiotin. The respective fractions obtained from both columns were combined and loaded onto a 15% sodium dodecyl sulfate-polyacrylamide gel for electrophoretic separation. The proteins within the gel were stained with Coomassie blue (BIO-RAD). The fraction with the highest amount of protein was selected and concentrated using centrifugal ﬁlter columns (RotiÕ_-Spin _MINI-10, _Carl _Roth, _Germany). _The protein concentration was determined using the Pierce

BCA Protein Assay kit (Thermo Scientiﬁc). The

HapR-N-Strep protein was stored at 20_{C in elution}

buffer containing 10% glycerol and 0.3 mM

Dithiothreitol (DTT). EMSA

The protocol for EMSA was established based on similar previously described protocols (12,13). Brieﬂy, DNA frag-ments were obtained through the PCR ampliﬁcation of gDNA from V. cholerae or plasmid DNA sequences using the primers listed in Supplementary Table S1. The amplicons were visualized on a 1.2% agarose gel. The

DNA fragments were (gel-) puriﬁed using either

GenEluteTM PCR Clean-Up or Gel Extraction kits (Sigma-Aldrich, Switzerland). The in vitro binding of HapR to the DNA probes was facilitated through incuba-tion of the puriﬁed protein with the DNA fragments in reaction buffer (10 mM Tris–HCl pH 8.0, 2.5% glycerol, 0.5 mM ethylenediaminetetraacetic acid, 100 mM KCl, 1 mM MgCl2and 2 mM DTT) for 20 min at 30C. After

the incubation, bromophenol blue and 2.5% glycerol were added to the reactions. The samples were loaded onto a pre-run (in 0.5 TBE) 8% polyacrylamide gel. The elec-trophoretic separation occurred for 1 h 40 min at 100 V. The proteins were transferred from the gel to a polyvinylidene diﬂuoride membrane for western blotting where indicated. All other gels were soaked for 30 min in a solution containing 0.2 mg/ml of ethidium bromide to stain

DNA. The pictures were captured with an UV

transilluminator (G:BOX, Syngene). The images were rotated, cropped and uniformly adjusted for brightness and contrast using Adobe Photoshop.

Quantitative reverse transcription PCR (qRT-PCR) The V. cholerae strains were grown for several hours in LB medium, as previously described (8). Where indicated, 0.02% arabinose was added to the medium to artiﬁcially induce natural competence (as TfoX is driven by the PBAD

promoter in this system) (8). The cell harvest, RNA prep-aration, and reverse transcription followed by quantitative PCR (qRT-PCR) (LightCyclerÕ _{Nano, Roche) were} per-formed as described earlier (8), except the cDNA was

prepared using the Transcriptor Universal cDNA

Master mix (Roche, Rotkreuz, Switzerland). The expres-sion values are given relative to the expresexpres-sion of the ref-erence gene gyrA (8). The gene-speciﬁc primers used for

qRT-PCR are indicated in reference (8) and in

(6)

RESULTS

The artiﬁcial expression of comEA increases natural transformation in a HapR-negative strain

Natural competence and transformation of V. cholerae occurs at a high cell density, as described earlier in the text. We previously showed that the expression of comEA is dependent on HapR, whereas many of the other competence genes, such as pilA, are expressed inde-pendently of this regulatory protein (8) (Figure 1). Based on this ﬁnding, we wanted to determine whether the transformability of a V. cholerae HapR mutant strain

could be rescued solely through mimicking the

HapR-mediated activation of comEA. Thus, strain hapR-TntfoX carrying the inducible plasmid pBAD-comEA was subjected to a chitin-independent natural transformation assay, as previously described (8). As shown in Figure 2, the artiﬁcial expression of comEA partly rescued the transformation phenotype in the hapR minus strain (lane 4). This result is consistent with that of a previous study showing that the natural transformation in HapR mutant strains could be partially rescued through deletion of dns, the gene repressed by HapR (7). However, it is unknown whether HapR or intermediate regulators directly inﬂuence comEA and dns expression.

A putative HapR binding site upstream of comEA is not essential for natural transformation

The direct binding of HapR to the promoter regions of the virulence regulator gene aphA and the gene encoding the hemagglutinin protease (HA protease) hapA has been demonstrated in vitro (12,13) (Figure 1). Furthermore, Tsou et al. (13) identified two distinct HapR-binding motifs, motif 1 and motif 2, using bioinformatics tools and experimental validation. We wanted to determine whether a similar motif would also be present in the upstream region of comEA, indicating the direct regula-tion of comEA through HapR. Indeed, we identified a DNA region located between 122 and 103 bp upstream of the comEA start codon, which highly resembled motif 1 (13) (Supplementary Figure S1A). To characterize the significance of this motif, we generated the plasmid pBR-[PWT]comEA, carrying comEA preceded by 200 bp of its

indigenous upstream region. This plasmid was used as a template for the site-directed mutagenesis of the left, right or both segments of the putative HapR binding site. The resulting plasmids, pBR-[Pmut L]comEA, pBR-[Pmut R]

comEA and pBR-[Pmut L/R]comEA, were assessed

for their ability to complement V. cholerae strains lacking the comEA gene on the chromosome in a chitin-independent transformation assay. As shown in Supplementary Figure S1, the transformation frequencies of all strains carrying mutations in the putative HapR binding site (panel B, lanes 4–6) were comparable with the frequency of the strain complemented with comEA preceded by its indigenous upstream region (lane 3). Similar results were obtained when natural competence was induced on a chitin surface using a previously described method (37). Indeed, in this chitin-dependent transformation assay, the complementation worked even

better (Supplementary Figure S1C). The most likely ex-planation for this difference in complementation efﬁciency is that the plasmids used in this study were derivatives of pBR322, which changes copy number under different growth conditions/growth rates (47). These results demon-strate that the putative HapR binding site is negligible for the regulation of comEA. However, it remains unknown whether HapR binds to a different sequence within the upstream region of comEA or whether HapR indirectly regulates this competence gene.

Direct and indirect regulation of competence genes by HapR

To determine whether HapR directly regulates certain genes that play a role in competence and transformation, we assessed the binding of HapR to the respective promoter regions of these genes in vitro. First, we puriﬁed the HapR protein using a tagged version of HapR through the cloning of plasmid pBAD-hapR-N-Strep (fusion between the sequence encoding for the Strep-tagII peptide and hapR). We then determined

whether hapR-N-Strep could complement the

transformation-negative phenotype of a V. cholerae hapR minus strain. As depicted in Supplementary Figure S2, the transformation frequencies of the strain hapR-TntfoX carrying hapR (lane 3) or hapR-N-Strep (lane 4) were almost identical, highlighting the in vivo functionality of this translational fusion.

Assured that HapR-N-Strep was fully functional in vivo, we heterologously expressed this gene in E. coli strain Origami 2TM (DE3) pLysS. The successful but low-level expression of HapR was veriﬁed through PAGE and western blot analyses (Supplementary Figure S2B). The N-terminally tagged protein was puriﬁed using a Strep-TactinÕ _{Sepharose column (see ‘Material and Methods’} section for details; Supplementary Figure S2C) and stored at 20_C.

Next, we characterized the in vitro binding of the purified HapR protein to specific DNA fragments using an EMSA. We first tested the in vitro binding ability of HapR to the comEA promoter region (the features of this

and other tested DNA probes are indicated in

Supplementary Tables S2–S4). As indicated in Figure 3A (lanes 5–8), HapR did not shift the comEA promoter (PcomEA)-speciﬁc DNA fragment, even at the highest

tested protein concentration (lane 8). The same result was obtained for another V. cholerae competence gene, comEC (21), whose expression is also inﬂuenced through HapR in vivo (8) (Figure 3A, lanes 9–12). In contrast, the promoter region of aphA, to which HapR binds in vitro (12), was successfully used as positive control in this assay (Figure 3A, lanes 1–4). We also examined the promoter of the competence gene pilA, which is not a component of the QS regulon in V. cholerae (8) (Figure 1). In this experi-ment, we used a longer DNA fragment of 600 bp, as this fragment was previously used to report pilA transcription using transcriptional reporter fusions (8). Consistent with the HapR-independent expression in vivo, HapR did not bind the pilA promoter in vitro (Figure 3B, lanes 5–8). A similar-sized fragment spanning the aphA promoter region

(7)

(aphA*) served as a positive control (lanes 1–4) to facili-tate the proper comparison with the PpilA probe. From

these data, we concluded that HapR neither binds to the comEA and comEC promoter regions nor to the pilA promoter region.

The other QS-regulated gene that plays a major role in natural transformation is dns (7) (Figure 1). To determine whether HapR directly or indirectly mediates the repres-sion of dns, we repeated the mobility shift assays using the dnspromoter region as a DNA template (Figure 4). An 250 bp fragment covering the region 203 bp to +48 bp from the start codon of dns was first tested (Figure 4A). Interestingly, HapR was able to shift the DNA fragment, although only at higher HapR protein concentrations than those sufficient to shift the positive control fragment (Figure 3A). We also included a non-related negative control in this assay [the recA promoter region; (8)], which did not show any mobility shift (data not shown). A western blot analysis of the dns-specific EMSA gel followed by the detection of HapR with a-HapR-specific antibodies confirmed that HapR protein co-localizes with the shifted DNA fragment (data not shown).

To identify the putative HapR binding site within the dnspromoter region, we PCR ampliﬁed several fragments around the dns start codon as indicated in Figure 4B (frag-ments A–H; details about the frag(frag-ments can be found in Supplementary Table S3). All fragments were tested for

HapR-dependent mobility shifts, and the results are indicated in the table on the left of Figure 4B. Apart from the outermost fragments (fragments A, G and H), all other probes remained bound to the HapR protein (Figure 4B). These fragments contained either the DNA sequence from 100 to 50 bp and/or the region from 50 to 1 bp, with respect to the start codon. Thus, we cloned these two sub-regions into vector pBR-Tet_MCSI (Table 1). The resulting plasmids were then used as PCR templates to obtain DNA probes of 200 bp in length (Figure 4B, fragments I and J). To ensure that HapR did not bind to the vector sequence, we also used a

DNA fragment solely derived from vector

pBR-Tet_MCSI as a negative control (Figure 4B, fragment K). The resulting EMSA patterns, after in vitro incubation with HapR, are depicted in Figure 4C. These data indicated that HapR binds to at least two separate HapR binding sites within the dns promoter in vitro. Although we did not ﬁnd any obvious HapR binding site for the dns sub-region of DNA fragment J, we identiﬁed a putative HapR binding site resembling motif 2 (13) within DNA fragment I (71 bp to 56 bp from the dns start codon but on the complementary strand; Supplementary Figure S3).

Despite both comEA and dns being HapR-dependent in vivo(8), the in vitro binding of HapR to the respective promoter region could only be demonstrated for dns. Thus, we concluded that HapR regulates comEA (and most likely also comEC) in an indirect manner.

A QS- and tfoX-dependent regulatory protein plays a major role in the natural competence and transformation of V. cholerae

Based on the results described earlier in the text, we hypothesized that an intermediary regulatory protein might exist between HapR and the QS-dependent

expres-sion of comEA. Based on previous microarray

expression-profiling data (2), we identified VC0396, which potentially encodes an intermediate protein in the competence regulatory network. VC0396 was initially annotated as a ‘transcriptional regulator of the LuxR family’ (42). Hereafter, we refer to this gene as QS and TfoX-dependent regulator (qstR). We validated the ex-pression of qstR in our chitin-independent experimental model using qRT-PCR. Comparable with other compe-tence genes, qstR was induced upon artificial low-level tfoX expression (Figure 5, highlighted results). This TfoX-dependent induction was not observed in a hapR-deficient background strain (Figure 5), which is consistent with the HapR-mediated regulation of qstR. Furthermore, a comparison of the expression pattern of qstR revealed high similarities with the expression pattern observed for comEA (Figure 5). Thus, these results confirmed our hy-pothesis that QstR could be an intermediary regulatory protein that signals between HapR and comEA.

The qRT-PCR data suggested the involvement of QstR in natural competence and transformation; there-fore, we deleted the qstR gene in the V. cholerae wild-type strain A1552 and in the strain A1552-TntfoX. The deletion of qstR did not result in any obvious

Figure 2. Artiﬁcial expression of comEA increases natural transform-ation in hapR negative strains. V. cholerae strains were tested for natural transformability through the artiﬁcial expressing of the trans-formation regulatory gene, tfoX, using 0.02% arabinose as inducer. Plasmid-encoded and PBAD-driven genes were simultaneously induced.

The tested strains were either A1552-TntfoX (WT-TntfoX, lane 1) or a hapRminus variant (hapR-TntfoX, lanes 2–4) all harbouring various plasmids. These plasmids were either the empty vector as control (lanes 1 and 2), plasmid pBAD-hapR (lane 3) or plasmid pBAD-comEA (lane 4). The natural transformation frequencies are indicated on the y-axis. The experiments were independently repeated three times, and the error bars reﬂect standard deviations. <d.l.: below detection limit (average d.l. of strain hapR-TntfoX was 2.9 109, as indicated with a dashed grey line). Statistically signiﬁcant differences were determined using Student’s t-test. *P < 0.05; for strain hapR-TntfoX, the value of the detection limit was used for statistical analysis.

(8)

competence-independent phenotype [e.g. growth in rich medium, colony morphology, chitin colonization (9), etc.]. We therefore examined whether the lack of qstR in-ﬂuences chitin-induced or chitin-independent natural transformation (Table 2). Indeed, for the qstR minus strain, transformation was below (chitin-dependent) or at the limit of detection (chitin-independent assay) and therefore reduced by at least four orders of magnitude compared with the respective parental strain (Table 2), highlighting the importance of QstR for natural

compe-tence and transformation. The extremely rare

transformants observed under chitin-independent competence-inducing conditions (Table 2) were most likely a reﬂection of low levels of comEA transcription even in the absence of QstR, whereas a comEA knockout strain was never transformable in our assay.

Next, we complemented the deletion strain by providing the qstR gene in trans, which restored natural transform-ation (Table 2). Consistent with our hypothesis that QstR acts as an upstream regulator of comEA (Figure 1), the exogenous expression of comEA also signiﬁcantly increased natural transformability in the qstR genetic background (Table 2). The transformation frequency in this setting (e.g. qstR/pBAD-comEA) was 10-fold higher than in strain hapR/pBAD-comEA (Figure 2), which is consistent with HapR-mediated repression of dns in the qstR mutant. To demonstrate that QstR acts downstream of HapR, we also artiﬁcially expressed qstR

from a plasmid in the hapR minus strain. Indeed, the exo-genous expression of qstR in trans in the hapR-TntfoX strain rescued natural transformation to the same extent as the hapR gene itself (Table 2). We therefore propose that HapR directly or indirectly inﬂuences qstR expression and that QstR regulates QS-dependent competence and transformation genes (Figure 1). Furthermore, as the overexpression of qstR in a hapR background apparently also alleviated the transformation barrier exerted by the Dns nuclease, we suggest that QstR might also be involved in the repression of dns (Figure 1).

HapR binds to the qstR promoter region

Next, we examined whether HapR regulates qstR in a direct or indirect manner. We repeated the in vitro HapR binding assay as described earlier in the text, using the upstream region of qstR as a probe. As indicated in Figure 6, HapR bound to the DNA fragment causing a shift in mobility.

To identify the location of the putative HapR binding site(s), we followed the strategy described for Figure 4, examining a plethora of DNA fragments derived from 359 bp upstream to +105 bp downstream of the start codon of qstR (Figure 6B). The probes (fragments A–G; details in Supplementary Table S4) were tested for in vitro binding to HapR (EMSA), as indicated in the ﬁgure. All shifted fragments shared the sequence spanning the region from 150 to 102 bp upstream qstR

Figure 3. HapR does not bind to the comEA, comEC and pilA promoters in vitro. EMSA using the comEA (PcomEA), comEC (PcomEC) (panel A) and

the pilA (PpilA) (panel B) upstream regions as a probe did not show any bandshift. The aphA promoter was used as a positive control (PaphA;

*indicates the longer fragment used in panel B as described in the text). A total of 40 ng (panel A)/80 ng (panel B) of DNA fragments were incubated without () or with increasing amounts of HapR-N-Strep protein, as schematized in the ﬁgure. L: DNA ladder (representative bp are indicated on the left). Solid arrow: unbound DNA probe. Dashed arrow: bound/shifted DNA.

(9)

(Figure 6B and Supplementary Table S4). Thus, we subcloned two fragments, i.e. the regions spanning either 150 to 120 bp or 150 to 102 bp upstream of the ATG start codon into vector pBR-Tet_MCSI (Table 1). The resulting plasmids were used as PCR templates to obtain linear DNA fragments of 200 bp in length (frag-ments H and I in Figure 6B). Fragment H, containing the

shorter stretch of the PqstRregion, did not exhibit in vitro

binding to HapR (Figure 6C). In contrast to fragment H, the 50 bp of the qstR upstream region within fragment I, were sufficient to facilitate HapR binding. On further exam-ination of this particular DNA region, we identified a sequence resembling in silico predicted motif 2 (13) [Figure 6D; note that the simplified consensus was slightly modified

Figure 4. The HapR protein binds to the dns upstream region. (Panel A) Binding of HapR to the upstream region of dns results in a shifted DNA fragment. Lanes 1–4: EMSA of the dns promoter fragment covering the region 203 to +48 bp with respect to the ‘‘ATG’’ start codon (corresponding to fragment C shown in panel B). The increasing amounts of HapR-N-Strep protein are depicted on the right of the image. (Panel B) Schematic represen-tation of the DNA region surrounding the dns start codon. The tested DNA fragments (A. to K.) spanning the respective region are depicted below the scheme. The dashed line in fragments I. to K. represents unrelated and plasmid-derived DNA. All fragments were tested for HapR-N-Strep-mediated in vitrobinding using EMSA, and the EMSA results are indicated in the left column. The large grey arrow depicts the dns gene (not to scale). (Panel C) HapR binding site(s) were associated with two 50 bp regions located within the dns promoter. DNA fragments (40 ng) of 200 bp length containing short parts of dns upstream region (100 to 50 bp for fragment I; 50 to 1 bp for fragment J) surrounded by unrelated and plasmid-derived DNA were subjected to EMSA using increasing amounts of HapR-N-Strep, as indicated in panel A. The negative control (fragment K) did not contain any Pdns-derived DNA sequence. L: DNA ladder. Solid arrow: unbound DNA probe. Dashed arrow: bound/shifted DNA.

(10)

from (13) to facilitate alignment with the putative motif upstream of qstR]. Indeed, after site-directed mutagenesis of two conserved bases within this motif (Figure 6D), the capability of HapR to shift the DNA fragment in the assay was lost (Figure 6C). Taken together, these results demon-strate that HapR directly binds to the qstR upstream region, and that this binding occurs in a sequence-speciﬁc manner, despite the fact that the motif does not perfectly match the in silicopredicted consensus (13).

QstR is essential for the TfoX-dependent induction of QS-dependent competence genes

The data provided earlier in the text indicated that HapR binds to the qstR promoter region. However, which com-petence genes does QstR subsequently regulate? To answer this question, we compared the expression levels of a plethora of competence and chitin metabolism genes, both under competence non-inducing and

competence-Table 2. QstR plays a major role in the natural transformation of V. cholerae

V. choleraestrain ±plasmid (gene on plasmid) Assay Transformation frequencyx _(±SD)y

A1552 Chitin-induced transform-ation [assayed on chitin

ﬂakes; (37)] 1.3 104(±5.7 105) - - - -qstR <d.l. (d.l.= 1.3 108) A1552-TntfoX Artiﬁcial and chitin-independent expression of tfoX (8) 2 104(±7.5 105)a = ** - - - -qstR-TntfoX 1.5 108(±3.6 109)a = ** - - - -A1552-TntfoX + 8.1 104(±6.2 104)b = **

(no gene; vector control)

- - - -qstR-TntfoX + 1.2 108(±9.1 109)b = **/c = **/d = **

- - - -qstR-TntfoX + (qstR) 4.4 105_{(±2.0 10}5₎c = **/e = **

- - - -qstR-TntfoX + (comEA) 1.8 107_{(±8.4 10}8₎d = **/e = **

- - - -hapR-TntfoX + <d.l. (d.l. = 3.1 109₎

- - - -hapR-TntfoX + (hapR) 3.9 105_{(±2.8 10}5₎f = n.s.

- - - -hapR-TntfoX + (qstR) 1.5 105_{(±9.2 10}6₎f = n.s.

x_{Average of at least three independent experiments.} y

Statistically signiﬁcant differences between transformation data indicated with the same characters were determined using Student’s t-tests. **P < 0.01.

n.s.= not signiﬁcantly different. <d.l.: below detection limit.

Figure 5. QstR is required for the induction of comEA and comEC. qRT-PCR data showing the expression of the indicated genes relative to gyrA in the wild-type background strain A1552-TntfoX and its hapR or qstR knockout derivatives (hapR-TntfoX and qstR-TntfoX). All three strains were grown under competence non-inducing (TntfoX ) and competence-inducing (TntfoX+) conditions. The highlighted results (shaded boxes) are first discussed in the text and indicate that the expression of qstR is TfoX- and HapR-dependent. The data represent the averages of three biological replicates. The error bars indicate standard deviations. Statistically significant differences were determined using Student’s t-tests. *P < 0.05, **P < 0.01, n.s. = not significant.

(11)

Figure 6. The HapR protein binds to the promoter region of qstR. (Panel A) HapR’s ability to bind to the upstream region of qstR was tested by EMSA. A DNA fragment covering the region 248 to 47 bp upstream the start codon of qstR (corresponding to fragment C shown in panel B) was used as a probe for the in vitro binding of HapR. The aphA promoter was used as a positive control. The concentration of HapR protein used in each lane is indicated above the images and schematized on the right of the figure. (Panel B) Scheme representing the DNA region, which surrounds the qstR start codon. DNA fragments (A. to I.) spanning regions upstream and within qstR are depicted below the scheme. A dashed line indicates the unrelated and plasmid-derived DNA of fragments H. and I. All fragments (A. to I.) were tested using EMSA, and the results are indicated in the left column. The large grey arrow depicts the qstR gene (not to scale). (Panel C) A HapR binding site was located within a 50 bp stretch upstream the qstRgene and site-directed mutagenesis abolished HapR’s ability to bind the qstR promoter. A total of 40 ng of DNA fragments (200 bp in length) containing short parts of the qstR upstream region (150 to 120 bp for fragment H; 150 to 102 bp for fragment I; and a mutated version thereof as indicated in panel D) surrounded by plasmid-derived and therefore qstR-unrelated DNA were subjected to EMSA. Only fragment I, containing the longer qstR upstream region, bound to HapR in vitro, resulting in a bandshift. The amounts of HapR-N-Strep used were as indicated in panel A. L: DNA ladder. Solid arrow: unbound DNA probe. Dashed arrow: bound/shifted DNA. Panel D: A HapR binding motif exists within the qstR promoter region. Simplified scheme of the in silico predicted HapR binding motif 2 [with slight modification from (13); e.g. four gaps () were introduced in the consensus sequence to allow proper alignment with the qstR promoter sequence]. A similar sequence located upstream of qstR (130 to 111 bp from the start codon) is illustrated in the middle row. This DNA sequence has been modified through site-directed mutagenesis (boxed residues), resulting in a mutated PqstRsequence as depicted in the lower row. The shadings indicate highly conserved (black background),

(12)

inducing conditions in three different strains. Apart from the tfoX-carrying transposon TntfoX (8) the genetic

back-grounds of these strains were either wild-type

(A1552-TntfoX), hapR-negative (hapR-TntfoX) or qstR-negative (qstR-TntfoX) (Figure 5). The results showed that the expression of comEA and comEC is sig-niﬁcantly lower under competence-inducing conditions in the two knockout strains compared with that observed in the corresponding wild-type strain, although the ex-pression levels were comparable between the two strains. In contrast, the competence-independent background ex-pression of comEA and comEC did not differ among any of the three strains tested (Figure 5). The results were completely different for dns. The expression of dns was signiﬁcantly reduced in the wild-type background strain upon tfoX induction, although the hapR transcript levels did not change (Figure 5). Furthermore, the expression

level of hapA, as a competence-unrelated but

HapR-dependent gene and therefore a direct reflection of HapR protein activity (Figure 1), did not change on competence induction. Consistent with HapR acting as repressor of dns, the dns expression levels were signifi-cantly increased in the hapR minus background strain, independently of the status of tfoX (induced or not) (Figure 5). For the newly identified regulator QstR, we observed a phenotype with respect to dns expression, which neither reflected the observations of the wild-type strain (e.g. TfoX-dependent repression of dns) nor those of the hapR minus strain (constitutively high expression of dns). In contrast, under competence-non-inducing condi-tions, dns was repressed through HapR in the qstR negative strain, but further repression on TfoX induction did not occur (Figure 5). We therefore speculate that QstR might exert additional repression on dns (as depicted with a dashed repression arrow in Figure 1). In contrast to the expression signatures of the QS-dependent genes comEA, comEC and dns (2,7,8), none of the other tested genes involved in competence [pilA, pilM, VC0047, dprA; (2,21)] or in the metabolism of chitin in V. cholerae [chiA-1, chiA-2, VCA0700; (9,14)] showed any significant difference between the three tested strains under

competence-inducing conditions (Supplementary

Figure S4).

DISCUSSION

We recently established a chitin-independent experimental model to study the regulatory pathway of natural compe-tence and transformation of V. cholerae (8). This assay is based on the artiﬁcial expression of low levels of the major regulator of transformation, TfoX, from a single chromo-somal copy (8). The advantages over competence induc-tion on chitin surfaces were as follows: (i) chitin sensing, chitin surface colonization and chitin degradation can be uncoupled from competence induction and (ii) the experi-ments were highly reproducible (see the ‘Results’ section) for the investigation of the events downstream of TfoX owing to a more homogeneous response within the bac-terial population (8). In this study, we used the system to identify a new regulatory protein involved in the natural

competence and transformation of V. cholerae, i.e. QstR, and determine some of its downstream target genes. We demonstrated that qstR expression is mediated through the two master regulators of QS and transformation, HapR and TfoX, respectively, thereby connecting these two pathways (Figure 1). Furthermore, using bacterial genetics and biochemical approaches, we provided evidence that QstR acts downstream of HapR and that HapR binds to a speciﬁc DNA sequence motif in the qstR promoter region. This motif resembles one of two in silico predicted HapR consensus binding sites (13), and site-directed mutagenesis abolished the in vitro binding of HapR to this promoter region. Finally and in accord-ance with its role downstream of HapR, we provided data demonstrating that QstR is only required for the regula-tion of a subset of competence genes. Indeed, consistent with previous data (8), we demonstrated here that the competence and chitin metabolism genes, apart from comEA and comEC, were fully inducible through TfoX even in the absence of HapR and QstR (Figure 5), including the genes chiA-1 and pilA, both of which were recently suggested by Antonova et al. (48) as positively

regulated through HapR. Apart from the

HapR-independent expression of pilA and chiA-1, we did not observe any in vitro binding of HapR to the pilA promoter, again suggesting that pilA expression occurs independently of HapR. We previously suggested that these discrepancies between studies might reflect the dif-ferences in the strains of V. cholerae O1 El Tor employed in different research laboratories (8). However, the data provided here are strongly supported by the results derived from yet another O1 El Tor strain, i.e. the first sequenced strain of V. cholerae N16961 (42). Microarray expression data derived after chitin-induction of strain N16961 showed the highly significant induction of pilA, chiA-1 and other competence- and chitin metabolism-related genes (14). Notably, the V. cholerae strain N16961 has a frameshift mutation within hapR, which abolishes QS (49,50) and consequently also natural trans-formation (2,23). Therefore, the expression data derived for this QS-defective strain N16961 are consistent with the HapR- (8) and QstR- (this study) independent expression of pilA, chiA and other competence and chitin metabolism genes (apart from comEA and comEC) within V. cholerae strain A1552 (23,36) used in this study.

The mechanism of how QstR regulates comEA, comEC and potentially dns remains unknown. Our preliminary data demonstrated that a tagged version of the QstR protein, which showed full functionality in vivo, did not bind to the comEA or comEC promoter region in vitro (data not shown). One possible explanation for this might be that instead of directly binding the comEA promoter,

QstR somehow enhances the activity of TfoX

(+/ CRP-cAMP) in vivo. However, the overexpression of qstRfrom a plasmid under non-competence-inducing con-ditions triggered comEA expression, even though not to the same level as observed under competence-inducing condi-tions (data not shown). Thus, we hypothesize that a (competence-speciﬁc) cofactor might be required for the full functionality of QstR in vivo and in vitro (Figure 1). QstR, together with its cofactor, might further repress

(13)

dns, as the qRT-PCR data suggest (Figure 5). The require-ment of a cofactor for the full activity of QstR is consist-ent with QstR being a LuxR-type regulatory protein. This annotation is based on the C-terminally located DNA-binding domain (LuxR_C_like [cd06170]; NCBI Conserved Domain database). According to the Pfam protein family database (51), ‘the LuxR-type binding helix-turn-helix (HTH) domain is a DNA-binding, HTH domain of about 65 amino acids. It is present in transcription regulators of the LuxR/FixJ family of response regulators’. This database further states that the HTH domain is typically located at the C-terminal region of the protein, whereas the N-terminal region of the protein often contains autoinducer binding, ligand binding or response regulatory domains. As dis-cussed earlier in the text, the autoinducer CAI-1 plays a pivotal role in natural competence and transformation of V. cholerae; therefore, it is tempting to speculate that QstR directly senses CAI-1. However, our preliminary data do not support this notion. We will further explore this hy-pothesis in the future.

VpsT is another prominent transcriptional regulator of the LuxR family in V. cholerae. Indeed, QstR and VpsT share 35% sequence identity but only within the C-terminal HTH domain. VpsT has been identified in the Yildiz group and is required for the expression of the Vibrio polysaccharide (vps) gene cluster (52,53). More precisely, VPS production in V. cholerae is tightly linked to the signalling pathway of the secondary messen-ger molecule cyclic di-GMP (c-di-GMP) (54,55). Indeed, it was shown that the regulatory protein VpsT inversely regulates motility and the VPS matrix production required for biofilm formation in V. cholerae owing to its ability to directly sense c-di-GMP (56). On the crystal-lization of VpsT, the authors identified a four-residue-long c-di-GMP binding motif (W[F/L/M][T/S]R) and provided evidence that the binding of c-di-GMP stabilizes the VpsT dimer. Krasteva et al. (56) also reported that ‘a subclass of VpsT and/or CsgD homologs exists with a proline substitution in position 3 (W[F/L/M]PR)’, which is indeed the case for QstR. Thus, it is unlikely that QstR binds c-di-GMP as a cofactor in a similar manner as VpsT.

In a recent study, Ferreira et al. (57) examined homologs of the LuxR/VpsT/CsgD family of transcriptional regulator and c-di-GMP signalling in another Vibrio species, V. parahaemolyticus. These authors compared the micro-array gene expression proﬁles of a wild-type and a mutant strain, with a deletion in the Scr system that affects the cellular levels of c-di-GMP (57). One of the few differentially regulated genes encoding a transcription factor was VP2710, a QstR homolog in V. parahaemolyticus. Indeed, based on a Basic Local Alignment Search Tool comparison, QstR and VP2710 share 52% identity (72% similarity), and GenoList (58) indicates VP2710 as the closest match of QstR in V. parahaemolyticus. Furthermore, qstR and VP2710 are both located in close proximity to the msh gene cluster (see later in the text), which is indicative of VP2710 playing a similar role in the regulation of natural compe-tence and transformation in this organism. The microarray data by Ferreira et al. (57) suggest that the expression of the

essential competence genes [e.g. homologs to the V. cholerae counterparts, which we identified through comparative genomics using GenoList (58)] was not altered under the tested experimental conditions, potentially reflecting that the main regulator of competence, TfoX, was not concomi-tantly expressed. Notably, natural transformation was only demonstrated in V. parahaemolyticus under chitin-inducing conditions (4) based on a previously published protocol established for V. cholerae (2). QstR homologues are also present in many other Vibrios such as Vibrio mimicus (ZP_06040190.1; 96% identity), Vibrio furnissii (VFA_000366/ZP_05876252.1; 71% identity), Vibrio anguillarum (VAA_01740/YP_004567366.1; 67% identity), Vibrio harveyi (VIBHAR_03706/YP_001446847.1; 52% identity) and Vibrio vulnificus (VV1_1429/AAO09869.1; 47% identity).

Basic Local Alignment Search Tool searches using the QstR amino acid sequence showed the partial annotation of these proteins as homologues of the CsgD family of LuxR-type transcriptional regulators. CsgD is a key regu-lator in Salmonella enterica and E. coli, required as a positive regulator for the expression of genes encoding curli fimbriae and the synthesis of the exopolysaccharide cellulose; both curli and cellulose are implicated in biofilm formation [for a review, see (59)]. Interestingly, in V. cholerae qstR (VC0396) is located in close proximity (though in opposite orientation) to the mannose-sensitive haemagglutinin pilus-encoding gene cluster (VC0398 to VC0414) (60,61). This type IV pilus participates in the initiation of biofilm formation on abiotic and biotic surfaces within the aquatic environment of V. cholerae [(62–64); for a review, see (65)]. Based on the genetic linkage between qstR and the msh cluster and the correl-ation between msh-dependent chitin colonizcorrel-ation (14,64) and chitin-induced competence (2), we examined the ex-pression of mshA, a gene encoding the major pilin subunit of the mannose-sensitive haemagglutinin pilus (66), under competence non-inducing and competence-inducing con-ditions (according to the experimental approach described for Figure 5). The relative expression of mshA was 57.4 (±1.2) and 63.7 (±13.7) in the wild-type background strain (A1552-TntfoX) and 63.0 (±17.6) and 63.6 (±20.8) in its qstR negative counterpart (qstR-TntfoX) under non-inducing and inducing conditions, respectively. No statistically significant differences were observed for these values between the strains and without or with tfoX expression (average of three biological replicates). Furthermore, we observed a strong chitin surface colon-ization phenotype (9) for an mshA mutant, i.e. absence of surface colonization accompanied by increased motility; this phenotype was not observed for the qstR mutant, which exhibited behaviours similar to the WT in this assay (data not shown). However, the hapR negative strain (hapR-TntfoX) showed a statistically significant increase of 2-fold in mshA expression compared with that of WT under both tfoX-non-inducing (relative expres-sion = 113.3 ± 7.5) and tfoX-inducing (100.5 ± 6.7) con-ditions. We conclude that HapR is involved in the regulation of the msh gene cluster and that this regulation is independent of QstR. Notably, Marsh and Taylor (61) speculated that the msh gene cluster has been horizontally

(14)

acquired as the ﬂanking genes are adjacent to each other in E. coli and as the region is ﬂanked by a 7-bp direct repeat. This could explain why qstR is located proximal to the msh cluster but is not involved in its regulation.

In summary, our study demonstrates that QstR is a new transcription factor involved in the regulatory circuitry of natural competence and transformation. We provide evidence that QstR is dependent on the QS regulator HapR, and that it acts downstream of HapR. The expres-sion of qstR is also dependent on tfoX expresexpres-sion. QstR is required for the downstream expression of competence genes, which we have previously shown to be QS regulated (e.g. comEA and comEC) (8). We also propose that QstR functions most efﬁciently in the presence of a cofactor, and future studies will elucidate the nature of such a po-tential factor.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR

Online: Supplementary Tables 1–4 and Supplementary Figures 1–4.

ACKNOWLEDGEMENTS

The authors would like to thank Sandrine Borgeaud and Cle´mentine The´voz for technical assistance. They also acknowledge Patrick Seitz for provision of V. cholerae strains and Juliane Ku¨hn for contributing primers to this work. They thank three anonymous reviewers for excellent suggestions on the manuscript.

FUNDING

Swiss National Science Foundation [31003A_127029 and 31003A_143356]. Funding for open access charge: Ecole Polytechnique Fe´de´rale de Lausanne (EPFL).

Conﬂict of interest statement.None declared. REFERENCES

1. Morens,D.M., Folkers,G.K. and Fauci,A.S. (2004) The challenge of emerging and re-emerging infectious diseases. Nature, 430, 242–249.

2. Meibom,K.L., Blokesch,M., Dolganov,N.A., Wu,C.Y. and Schoolnik,G.K. (2005) Chitin induces natural competence in Vibrio cholerae. Science, 310, 1824–1827.

3. Gulig,P.A., Tucker,M.S., Thiaville,P.C., Joseph,J.L. and Brown,R.N. (2009) USER friendly cloning coupled with chitin-based natural transformation enables rapid mutagenesis of Vibrio vulniﬁcus. Appl. Environ. Microbiol., 75, 4936–4949. 4. Chen,Y., Dai,J., Morris,J.G. Jr and Johnson,J.A. (2010) Genetic

analysis of the capsule polysaccharide (K antigen) and exopolysaccharide genes in pandemic Vibrio parahaemolyticus O3:K6. BMC Microbiol., 10, 274.

5. Pollack-Berti,A., Wollenberg,M.S. and Ruby,E.G. (2010) Natural transformation of Vibrio ﬁscheri requires tfoX and tfoY. Environ. Microbiol., 12, 2302–2311.

6. Blokesch,M. and Schoolnik,G.K. (2007) Serogroup conversion of Vibrio choleraein aquatic reservoirs. PLoS Pathog., 3, e81. 7. Blokesch,M. and Schoolnik,G.K. (2008) The extracellular nuclease

Dns and its role in natural transformation of Vibrio cholerae. J. Bacteriol., 190, 7232–7240.

8. Lo Scrudato,M. and Blokesch,M. (2012) The regulatory network of natural competence and transformation of Vibrio cholerae. PLoS Genet., 8, e1002778.

9. Blokesch,M. (2012) Chitin colonization, chitin degradation and chitin-induced natural competence of Vibrio cholerae are subject to catabolite repression. Environ. Microbiol., 14, 1898–1912. 10. Seitz,P. and Blokesch,M. (2012) Cues and regulatory pathways

involved in natural competence and transformation in pathogenic and environmental Gram-negative bacteria. FEMS Microbiol. Rev., September 20 (doi: 10.1111/j.1574–6976.2012.00353.x; epub ahead of print).

11. Ng,W.L. and Bassler,B.L. (2009) Bacterial quorum-sensing network architectures. Annu. Rev. Genet., 43, 197–222. 12. Kovacikova,G. and Skorupski,K. (2002) Regulation of

virulence gene expression in Vibrio cholerae by quorum sensing: HapR functions at the aphA promoter. Mol. Microbiol., 46, 1135–1147.

13. Tsou,A.M., Cai,T., Liu,Z., Zhu,J. and Kulkarni,R.V. (2009) Regulatory targets of quorum sensing in Vibrio cholerae: evidence for two distinct HapR-binding motifs. Nucleic Acids Res., 37, 2747–2756.

14. Meibom,K.L., Li,X.B., Nielsen,A.T., Wu,C.Y., Roseman,S. and Schoolnik,G.K. (2004) The Vibrio cholerae chitin utilization program. Proc. Natl Acad. Sci. USA, 101, 2524–2529. 15. Yamamoto,S., Morita,M., Izumiya,H. and Watanabe,H. (2010)

Chitin disaccharide (GlcNAc)2induces natural competence in

Vibrio choleraethrough transcriptional and translational activation of a positive regulatory gene tfoXVC_{. Gene, 457, 42–49.}

16. Yamamoto,S., Izumiya,H., Mitobe,J., Morita,M., Arakawa,E., Ohnishi,M. and Watanabe,H. (2011) Identiﬁcation of a chitin-induced small RNA that regulates translation of the tfoX gene, encoding a positive regulator of natural competence in Vibrio cholerae. J. Bacteriol., 193, 1953–1965.

17. Cameron,A.D. and Redﬁeld,R.J. (2006) Non-canonical CRP sites control competence regulons in Escherichia coli and many other gamma-proteobacteria. Nucleic Acids Res., 34, 6001–6014. 18. Cameron,A.D. and Redﬁeld,R.J. (2008) CRP binding and transcription activation at CRP-S sites. J. Mol. Biol., 383, 313–323.

19. Sinha,S., Cameron,A.D. and Redﬁeld,R.J. (2009) Sxy induces a CRP-S regulon in Escherichia coli. J. Bacteriol., 191, 5180–5195. 20. Sinha,S., Mell,J.C. and Redﬁeld,R.J. (2012) Seventeen

Sxy-dependent cyclic AMP receptor protein site-regulated genes are needed for natural transformation in Haemophilus inﬂuenzae. J. Bacteriol., 194, 5245–5254.

21. Suckow,G., Seitz,P. and Blokesch,M. (2011) Quorum sensing contributes to natural transformation of Vibrio cholerae in a species-speciﬁc manner. J. Bacteriol., 193, 4914–4924. 22. Antonova,E.S. and Hammer,B.K. (2011) Quorum-sensing

autoinducer molecules produced by members of a multispecies bioﬁlm promote horizontal gene transfer to Vibrio cholerae. FEMS Microbiol. Lett., 322, 68–76.

23. Blokesch,M. (2012) A quorum sensing-mediated switch contributes to natural transformation of Vibrio cholerae. Mob. Genet. Elements, 2, 224–227.

24. Schauder,S., Shokat,K., Surette,M.G. and Bassler,B.L. (2001) The LuxS family of bacterial autoinducers: biosynthesis of a novel quorum-sensing signal molecule. Mol Microbiol, 41, 463–476. 25. Miller,M.B., Skorupski,K., Lenz,D.H., Taylor,R.K. and

Bassler,B.L. (2002) Parallel quorum sensing systems converge to regulate virulence in Vibrio cholerae. Cell, 110, 303–314. 26. Higgins,D.A., Pomianek,M.E., Kraml,C.M., Taylor,R.K.,

Semmelhack,M.F. and Bassler,B.L. (2007) The major Vibrio choleraeautoinducer and its role in virulence factor production. Nature, 450, 883–886.

27. Lenz,D.H., Mok,K.C., Lilley,B.N., Kulkarni,R.V., Wingreen,N.S. and Bassler,B.L. (2004) The small RNA chaperone Hfq and multiple small RNAs control quorum sensing in Vibrio harveyi and Vibrio cholerae. Cell, 118, 69–82.

28. Lenz,D.H. and Bassler,B.L. (2007) The small nucleoid protein Fis is involved in Vibrio cholerae quorum sensing. Mol. Microbiol., 63, 859–871.

29. Seper,A., Fengler,V.H., Roier,S., Wolinski,H., Kohlwein,S.D., Bishop,A.L., Camilli,A., Reidl,J. and Schild,S. (2011)